mort96

19 ms·

421.
by mort96 6mo ago
Sounds like a job for the Linux Foundation maybe? You don't need anyone's permission to make a distro, that's true, but if you notify Debian, Canonical, Fedora, Red Hat and Arch you're covering a very large fraction of u
422.
by mort96 6mo ago
Oh shit I had assumed PyTorch Lightning was affiliated with PyTorch. Not a great name for an unaffiliated third party thing.
423.
by mort96 6mo ago
How about fd of the file you wanna rename, dirfd of the directory you want to open it in, and name of the new file? You could then represent a "rename within the same directory" as: dfd = opendir(...); fd = openat(dfd, "a&quo
424.
by mort96 6mo ago
But the point is, there is no line which separates white and black (or green and blue). 50% grey is neither black nor white, it's grey. Turquoise is neither green nor blue, it's turquoise.
425.
by mort96 6mo ago
> Ten years ago most scientists would probably have agreed that The Turing Test was sufficient but the goalposts shifted when ChatGPT passed that. The popularity of, and lack of consensus on, the Chinese room thought experiment kind of i
426.
by mort96 6mo ago
> What are you doing when you are not outputting tokens? You have a thought, evaluate it, refine it, repeat. I mean I do think sometimes even when not typing? > Will this lead to AGI? That’s still TBD. This is literally what I have be
427.
by mort96 6mo ago
And it does this thinking without producing tokens?
428.
by mort96 6mo ago
I've seen it in a couple software developer contracts here in Norway. I find it despicable and have always gotten it removed from any contract before signing. I don't get why it's even legal to have in contracts. I certainly
429.
by mort96 6mo ago
I don't think you showed counter examples? Or can you link me to a paper which describes a language model thinking without predicting tokens?
430.
by mort96 6mo ago
> If what you are saying is true, then LLMs wouldn't be able to handle out-of-distribution math problems without resorting to tool use. Yet they can. When you ask a current-generation model to multiply some 8-digit numbers, and forb
431.
by mort96 6mo ago
> Inability to introspect your own word selections does not mean it’s meaningfully different from what an LLM does. There is plenty of evidence that humans do a lot of things that are not driven by conscious choice and we rationalize it
432.
by mort96 6mo ago
> (Edit: to answer your question, the substrate your mind uses to drive this process can be considered an array of tokens that, themselves, can be considered 'words.') This seems like a factual claim. Can you link a source? (Al
433.
by mort96 6mo ago
What's your argument? An fMRI can tell which finger I will lift first before that information makes its way to my consciousness, ergo next word prediction is sufficient for general intelligence? Do you hear yourself?
434.
by mort96 6mo ago
I don't understand what part of what I said you disagree with.
435.
by mort96 6mo ago
Before I start typing, I think abstractly about the topic and decide on what I shall write in response. Due to the linear nature of time, typing necessarily happens one word at a time, but I am never producing a probability distribution of
436.
by mort96 6mo ago
I'm not sure why you think you know the human brain works through predicting the next token. It's not supernatural, I believe that an artificial intelligence is possible because I believe human intelligence is just a clever arra
437.
by mort96 6mo ago
I'm sorry but the input to a model is a sequence of tokens and the output is a probability distribution of what's the most likely next token. It's a very very very fancy next token predictor but that is fundamentally what it
438.
by mort96 6mo ago
If I'm reading you right, your opinion is essentially: "If building bigger and bigger statistical next word predictors won't lead to artificial general intelligence, we will never see artificial general intelligence" I d
439.
by mort96 6mo ago
What you're saying is true for phones too, yet phone manufacturers lock down their hardware to only allow it to run Google's operating system. The argument that "hardware manufacturers live by selling hardware and don't
440.
by mort96 6mo ago
Do we have any proof that those 80m 17s didn't include searching the Internet?
441.
by mort96 6mo ago
I mean the algorithms for that are known, and PipeWire/PulseAudio implements them competently. But it still has to be done.
442.
by mort96 6mo ago
I mean so could any other hardware vendor? No Mac in history has been locked down in the way you describe, and there's really no indication that Apple would start now. If they were ever going to, the ARM transition would've been t
443.
by mort96 6mo ago
I'm not really sure what it would mean for M1 air/pro to work better at this point to be honest, other than I guess power consumption during sleep but that's supposedly a super tricky problem that can't be "solved&q
444.
by mort96 6mo ago
The huge advantage of x86_64 isn't that it's a stable platform, but that the big hardware vendors maintain their own Linux driver. Nobody needs to reverse engineer how AMD GPUs work, you can just use AMD's driver. Nobody need
445.
by mort96 6mo ago
The way mainline kernel development happens is that people work in their own fork and then send patches upstream, just like what Asahi is doing. The only thing that makes Asahi special is that there's so much weird bespoke hardware
446.
by mort96 6mo ago
Eh no, sorry. The practical result is that a ton of people who have absolutely no interest in U2 has Songs of Innocence start playing when they don't want it. It plays when people turn on their cars. It plays when people connect to Blu
447.
by mort96 6mo ago
I feel like that's the kind of thing it's easy to not recognise as a terrible idea until after it's done, because so much of what makes it a bad idea is a consequence of the rest of the system. Imagine if everything else su
448.
by mort96 6mo ago
Then why do I still see USB 3.2 generation 2x2
449.
by mort96 6mo ago
USB 3.2 used to be what we now call USB 3.2 gen 2x2, doesn't it? So it used to be that the version dictated the max speed: 3.0 was 4Gb/s, 3.1 was 10, and 3.2 20, right? But then they decided to memory hole that and now USB 3.0 a
450.
by mort96 6mo ago
No, we haven't. You can start at any byte in a UTF-8 document and resume reading coherent text. If you start reading from the middle of a multi code point sequence, then the first couple of glyphs may be wrong, for example you may see

More ›